CLIN.CHEM.24/6,900-904 (1978)

Heparin-Mn2

Quantitationof High-Density-LipoproteinCholesterol:

An UltrafiltrationProcedure for Lipemic Samples G. Russeli Warnick and John J. Albers

We describe a modified heparin_Mn2+ procedure for high-density-lipoprotein cholesterol quantitation, especially in lipemic samples. High-density-lipoproteins may be estimated as cholesterol remaining in plasma supernates after precipitation of other lipoproteins by heparin and

Mn2+ treatment. However, in lipemic samples or those from non-fasting individuals, the lower density of the preCipitated chylomicrons, very-low-, and Iow-density-iipoproteins frequently prevents their sedimentation by the usual low-speed centrifugation, and high-density-lipoprotein cholesterol thus is overestimated in the resulting turbid supernates. Sedimentation is improved by a twofold increase in Mn2+ concentration to 92 mmol/liter. The procedure reported here produced clear supernates in more than 95% of sampies tested. Any remaining turbid supernates can be cleared by a simple, convenient ultra-

additional proteins such as fibrinogen in plasma may interfere with the heparin-Mn2 precipitation of non-HDL lipoproteins. We observed incomplete precipitation of the apoBassociated lipoproteins from EDTA-treated plasma by the original heparin-Mn2 method with Mn2 at 46 mmol/liter, final concentration, which was remedied by a twofold increase in Mn2 to 92 mmol/liter (9). This increase in Mn2 was incorporated into a method with other modifications for improved accuracy and convenience. Heparin and Mn25 were

added to plasma in a combined

reagent,

decreasing

the pi-

petting manipulations and with volumes better suited for use with manual pipets. Incubation time was reduced to 10 mm at room temperature before centrifugation at 1500 X g for 30 minat4 #{176}C(9). With lipemic samples or samples from non-fasting subjects, the density of the chylomicron-VLDL-LDL fraction precipitated with heparin-Mn2 frequently is too low for it to filtration technique. The filtration removed essentially all sediment. These lipoproteins thus remain suspended in the of the very-low- and low-density-lipoproteins without resolution, producing obvious turbidity. Any visible supernatant moving appreciable amounts of high-density-iipoproturbidity immediately after centrifugation indicates incomteins. plete removal of the non-HDL lipoproteins-and hence overestimation of HDL. [After overnight storage at 4#{176}C, a MnO The inverse relationship between HDL’ concentration and precipitate develops and after three to four days some HDL may precipitate (9)1. At the higher Mn2 concentration (92 cardiovascular disease risk was first identified in 1951 (1). More recently, epidemiological studies have emphasized the mmol/liter), precipitate density is increased relative to soluimportance of HDL as a negative risk factor (2-4). As a result, tion density, facilitating sedimentation. In one set of 89 clinical laboratories have experienced an increased demand samples, 5% of the supernates were turbid by the modified for quantitation of this lipoprotein. A method in which hepprocedure as compared to 10% after precipitation with 46 arm and Mn2+ are used to precipitate the apoB-containing mmol of Mn2 per liter (9). In another group of 375 samples, lipoproteins, primarily VLDL and LDL, with estimation of 8.5% were turbid at 46 mmol of Mn2+ per liter, as compared HDL in terms of the cholesterol remaining in the supernate, to 2.5% by the modified procedure (10). was described for serum and subsequently applied to Three procedures have been described for circumventing EDTA-treated plasma (5-7). EDTA-treated plasma is preturbid supernates in lipemic samples. Turbid supernates may ferred to serum for lipid and lipoprotein determinations for be cleared by centrifugation at a higher speed; 12 000 X g for the following reasons. For clot formation, blood must stand 10 mm (Method B) usually causes the unsedimented lipoat room temperature, which may produce changes in lipid! proteins to layer over the clear subnatant solution (9). Chylipoprotein distribution. Chylomicrons may be trapped in the lomicrons and VLDL can be removed by 20-h ultracentrifuclot and so removed from serum during centrifugation. Serum gation at density 1.006 kg/liter before heparmn-Mn2 treatcontains heavy metals that are known to promote auto#{246}x- ment (Method C) (7). A disadvantage of both these procedures idation of the lipids, whereas EDTA in plasma chelates metal is that an ultracentrifuge is required, and additionally for ions, preventing auto#{246}xidation(7, 8). However, EDTA or the Method C, the time and manipulations required to obtain the d >1.006 fraction. A third approach (Method D) is to dilute the lipemic sample with isotonic saline solution before adding Northwest Lipid Research Clinic, Harborview Medical Center, Seattle, Wash. 98104; and Department of Medicine, University of heparin and Mn2+ reagents, thus decreasing the solution Washington, Seattle, Wash. density and facilitating sedimentation (5). The dilution, ‘Nonstandard abbreviations used: apoA-I, apolipoprotein A-I, the however, also proportionately reduces the already-low HDL principal protein of HDL; apoB, apolipoprotein B, the major protein cholesterol value and may magnify technical error. of LDL and VLDL; EDTA, disodium ethylenediaminetetraacetate; We have investigated an alternative method (Method A), HDL, high-density-lipoprotein; LDL, low-density-lipoprotein; and using ultrafiltration to remove the supernatant turbidity asVLDL, very-low-density-lipoprotein. sociated with lipemic samples (11). Various ultrafilters and Received Feb. 6, 1978; accepted Mar. 27, 1978. 900

CLINICAL CHEMISTRY,

Vol. 24, No. 6, 1978

prefilters

were tested,

to determine

the most suitable

com-

bination for specific removal of the apoB-associated lipoproteins, primarily VLDL and LDL, without retention of HDL. Lipoprotein cholesterol and triglyceride concentrations in filtrates were compared to those in supernates obtained by centrifugation. ApoA-I, the major protein of HDL, was measured by radial immunodiffusion assay. Removal of the apoB-associated lipoproteins was checked by an radial immunodiffusion assay in which antibody to apoB was used. The ultrafiltration approach to HDL quantitation in lipemic samples was compared to the other three methods-B, C, and D-in terms of cholesterol, triglyceride, and apoB-associated cholesterol in filtrate or supernate.

Materials and Methods

through

filters.

Cholesterol,

in the filtrates filtered

0.22-, 0.45-, and 0.8-nm filters without preapoA-I, and triglyceride concentrations were compared to those in the remaining un-

supernatant

Second,

solutions.

heparin_Mn2+

treated

plasma

samples

were fil-

tered without centrifugal removal of the apoB-associated lipoproteins to determine the best filter for their specific removal. Heparmn-Mn2 solution (0.7 ml) was added to 7.0-ml plasma samples and thoroughly mixed. Two-milliliter portions were filtered through 0.22-, 0.45-, and 0.8-am filters without prefilters. The apoB-associated lipoproteins were removed from the remaining unfiltered heparmnMn2+treated plasma by centrifugation. HDL recovery was determined by comparing cholesterol and apoA-I concentrations in filtrates to

those in the corresponding supernates. LDL/VLDL removal was determined by measuring the apoB-associated cholesterol

Samples

in filtrates.

Plasma samples were obtained from normal and hyperlipidemic subjects according to the Lipid Research Clinic Protocol (7). Blood was collected from the antecubital vein into 15-mi Vacutainer Tubes containing 22.5 mg of dry disodium EDTA (3218-XF282-3875; Becton-Dickinson, East Rutherford, N.J. 07073). Immediately after thorough mixing, samples were cooled to 4 #{176}C and within 2 h cells were removed by centrifugation.

Heparin-Mn2

Precipitation

The apoB-associated lipoproteins were precipitated from plasma by heparin_Mn2+ treatment as described previously (9). The following method is designated the “modified procedure” (Mn2+ concentration, 92 mmol/liter). Prepare a 1.06 mol/liter solution of MnCl2-4 H20. Add 0.6 ml of sodium heparin (Lipo-Hepin, 40 X 106 USP units/liter, about 280 g/liter; Riker Laboratories, Northridge, Calif. 91324) to 10,0 ml of 1.06 mol/liter MnC12 solution. This combined heparin_Mn2+ solution, stored at 4 #{176}C, is stable for at least a month. Heparin from other sources may also be suitable, but each lot should be tested to determine if precipitation is complete with this concentration. To accommodate heparin preparations of different initial concentration, adjust the concentration of the original MnCl2 solution to obtain a Mn2 concentration of 1.0 mol/liter and a heparin concentration of about 16 g/liter in the combined solution. Add 0.2 ml of the combined heparin_Mn2+ solution to 2.0 ml of plasma, with thorough mixing. Allow the samples to stand for 10 mm at room temperature, then centrifuge at 1500 X g for 30 mm at 4 #{176}C. Aspirate the clear supernates for analysis. Filter any turbid supernates through a 0.22-pm filter protected by two depth prefilters2 assembled as follows: Place a 0.22-tim filter, 25 mm in diameter, on the lower support of a 25-mm Swinnex filter holder. Place the silicone gasket on the filter and fit an

AP 15 and an AP 20 glass depth prefilter,

each 22 mm in di-

ameter,

upper

Swinnex Bedford,

filtered

within

the

gasket.

Tighten

the

unit

of the

holder over the filter assembly (Millipore Corp., Mass. 07130). Force the turbid supernate through

the filter assembly with moderate positive pressure from a syringe and collect the clear filtrate for analysis. This filter combination was selected after the following experiments. First, clear supernates were filtered to determine if HDL was removed by ultrafiltration. Normo-lipidemic plasma samples (7.0 ml) each were treated with the combined heparin-Mn2 solution (0.7 ml) and the apoB-associated lipoproteins were removed by centrifugation at 1500 X g for 30 mm. Two-milliliter portions of the supernatant solutions were 2 A “depth” filter retains throughout its fibre matrix, in contrast to a “micron” filter, which retains particles exceeding a certain size only (essentially) on its surface.

Third, similar experiments were done except that two glass depth prefilters, an AP 15 and an AP 20, were added to prevent plugging of the ultrafilters. In the fourth series of experiments, the ultrafiltration procedure (Method A) was compared to methods B, C, and D for HDL cholesterol quantitation in hypertriglyceridemic samples.

Samples

were treated

with the combined

heparin-

Mn2 solution as explained previously. Method A: An aliquot of each turbid supernate was filtered with the 0.22-Mm filter with an AP 15 and an AP 20 prefilter assembled as above. Method B: A second aliquot was subjected to centrifugation at 12 000 X g for 10 mm in a SW 41 rotor (Beckman Instruments, Palo Alto, Calif. 98052). The clear subnatant solution was obtained

by aspirating

with a 23-gauge

the side of the tube (9). Method

needle

through

C: Heparmn-Mn2 (0.2 ml) performed on 2.0 ml of the

precipitation was simultaneously d >1.006 fraction obtained previously from each sample. Five milliliters of plasma was overlayered with 1.5 ml of 0.15 mol/ liter NaCl in a cellulose nitrate tube for the 40.3 rotor (Beckman Instruments) and centrifuged at 37 000 rpm for 20 h at 10 #{176}C. The d >1.006 fraction was recovered by a tube slicing technique and readjusted to the original plasma volume (7). Method D: A 1.0-mi portion of the lipemic plasma was diluted with 1.0 ml of 0.15 mol/liter NaCI solution before 0.2 ml of the combined heparin_Mn2+ solution was added. In all methods samples were incubated for 10 mm at room temperature before filtration or centrifugation, and in Methods C and D precipitates were sedimented by centrifugation at 1500 X g for 30 mm.

In the original heparmn-Mn2 precipitation procedure, Mn2+ was at 46 mmol/liter final concentration, half that used in these experiments (7). We performed a parallel series of experiments to validate the ultrafiltration approach for removal of precipitates at the lower Mn2+ concentration. In this method, designated “46 mmol/liter Mn2,” heparin (final concentration, 1.3 g/liter) and Mn2+ solutions were added to plasma sequentially, with mixing, followed by incubation at 4 #{176}C for 30 mm before centrifugation or filtration, which were done as described above for the modified procedure. Methods A-D were compared in hypertriglyceridemic samples after precipitation with 46 mmol of Mn2 per liter.

Lipid Analysis Cholesterol and triglycerides were quantitated by Lipid Research Clinic continuous-flow (AutoAnalyzer II) procedures (7). Supernates and filtrates were mixed thoroughly to resuspend any precipitates before extraction into isopropanol in the presence of zeolite mixture. Cholesterol was determined by a Liebermann-Burchard reagent method and triglyceride by a fluorometric 2,4-pentanedione procedure. Standard solutions and quality control samples were provided by the Lipid CLINICAL CHEMISTRY, Vol.24.No. 6, 1978

901

Table 1. Recovery of Cholesterol, Triglycerides, and ApoA-l from 10 Clear Heparin-Mn2 liter) Supernates after Ultrafiltration Mean concentrations, Cholesterol

(92 mmol/

mg/liter

Triglycerides

ApoA-I

Unfiltered supernates

545

115

1318

0.22-zm filtrate

517 (23)b

162 (48)a.c

1264 (77)5

0.45-zm filtrate 0.8-jm filtrate

527(16)b

177(59)b

1275 (62) 1269(64)C

526(17)b

118(11)

a Number in parentheses is the paired standard deviation of the difference between concentrations In the filtrate and the corresponding supernate (difference2/2n)”t. b

C

Paired Student’s f-test P < 0.005. P < 0.05

Standardization Laboratory of the Center for Disease Control, Atlanta, Ga. 30333. CV’s of 0.6, 0.9, and 1.4% were obtained for CDC pools (Q6, Q3, and MQ-2) with cholesterol concentrations of 2760, 1540, and 690 mg/liter, respectively, analyzed four times per daily run for 12 months. The corresponding triglyceride CV’s for Q6 (2480 mg/liter) and Q3 (690 mg/liter) were 1.1 and 3.0%, respectively. Mean values for these pools were within 20 mg/liter of Lipid Standardization Laboratory reference target values. Analysis conditions were designed to maximize the measurement precision in the supernates and filtrates that have low HDL concentrations. To eliminate between-run and between-tray variation, comparison samples from each subject were analyzed consecutively in one tray. Systematic bias effects of within-tray baseline drift and scale expansion were minimized by randomizing the order of the comparison samples for each subject. Sample interaction (carryover) was minimized by analyzing samples in duplicate, sequentially, and reporting only the second value. Instrument response did not deviate significantly from linearity, even at very low concentrations. Serial dilutions to 1:16 of a 650 mg/liter heparmn-Mn2 supernate were all within 10 mg/liter of the expected values. Lipid results from the continuous-flow analyzer were corrected for the dilution caused by the added heparin_Mn2+ reagent and the NaCI solution.

#{176}C for 72 h in a humidity chamber before precipitation ring diameters were measured. A standard was prepared by pooling plasma d >1.006 fractions (with NaN3, 500 mg/liter, as preservative) and quantitating LDL cholesterol as total cholesterol minus HDL cholesterol. Concentrations of five dilutions of this standard ranging from 10 to 200 mg of apoB-associated cholesterol per liter, prepared freshly for each assay, were linearly related to precipitation ring diameter squared. The antibody concentration in the gels was adjusted to provide sensitivity to about 5 mg of LDL cholesterol per liter.

Results and Discussion

Portions of 10 clear (free of apoB-associated cholesterol) heparin_Mn2+ supernates, obtained from normolipidemic plasma by the modified procedure, were filtered through 0.22-, 0.45-, and 0.8-sm filters without prefilters (Table 1). Cholesterol in the supernate was decreased slightly but significantly from a mean of 545 mg/liter to 517, 527, and 526 mgI liter on use of the 0.22-, 0.45-, and 0.9-Mm filters, respectively. A corresponding decrease was observed for ApoA-I, from a mean of 1318 mg/liter to 1264, 1275, and 1269 mg/liter, respectively. This slight decrease in HDL, approximately 3-5%, is probably a result of adsorption onto the ultrafiltration membranes. The slight decrease is statistically significant because of the excellent precision in the cholesterol analyses. Apolipoprotein Quantitation The triglyceride concentrations were increased significantly ApoA-I was quantitated in supernate and filtrate solutions with the 0.22- and 0.45-Mm filters, from a mean of 115 mg/liter by a radial immunodiffusion method (12). Samples were dito 162 and 177mg/liter, respectively. This increase in apparent luted with an equal volume of tetramethylurea (Burdick and triglyceride, was attributable to a non-triglyceride fluorescent Jackson Laboratories, Muskegon, Mich. 49942), further dicontaminant leaching from some lots of filters into the filtrate, luted five-fold with tris(hydroxymethyl)ammnomethane buffer and was also observed in de-ionized water after 0.22-sm fil(10 mmol/liter, pH 8.0) containing 8 mol of urea per liter and tration. Because HDL is generally quantitated in terms of its incubated at 23 #{176}C for 30 mm. After incubation samples were cholesterol moiety, the artifactual increase in triglyceride is applied to 1.8-mm wells in gels of 1.0% agarose (Sigma only a concern in studies of HDL composition. Presumably, Chemical Co., St. Louis, Mo. 63178), containing, per liter, 20 only fluorometric triglyceride methods would be affected. mmol of tris(hydroxymethyl)aminomethane buffer, pH 8.0, Similar results were obtained for filtrates of three super0.15 mol of NaCI, 500mg of EDTA, 500mg of NaN3, 5.0 g of nates obtained after precipitation at 46 mmol of Mn2 per liter. Filtrate cholesterol concentrations averaged 1% and bovine serum albumin (Sigma), and rabbit antibody specific averaged 4% less than the original for human apoA-I. The gels were incubated for 72 h at 37 #{176}C apoA-I concentrations in a humidity chamber. Precipitation ring diameters were values for the supernates. measured with a calibrated viewer (Transidyne General Corp., Portions of 16 plasma samples after heparin-Mn2 addition Ann Arbor, Mich. 48106). ApoA-I standards were purified (92 mmol of Mn2 per liter), but without centrifugation, were from human HDL and standardized against albumin as defiltered through 0.22-, 0.45-, and 0.8-tim filters without depth scribed previously (12). Concentrations of six standards prefilters. Filtration rates through the 25-mm diameter filters ranging from 500 to 2500 mg/liter were linearly related to with moderate pressure from a 5-ml syringe were low,